StraataNightshift Guides

Straata / Articles

AI anxiety for software engineers: what to do when coding tools keep getting better

Worried AI will replace software engineers? Sort your tasks, test the tools on your own codebase, build a review habit and a 90-day plan.

From The Replacement Fear · Published by Nightshift Guides · · 5 min read

A teammate ships a feature in an afternoon with an assistant. Your company announces an "AI-first" push. A thread on a forum says junior roles are vanishing. If you write software for a living, these moments can leave a steady question running in the background: what is my job going to be?

This article does not predict the future of the profession. Nobody can do that reliably. It offers a way to turn a vague worry into something you can examine, using your own recent work as the evidence.

Why this hits engineers in a particular way

Many engineers build their identity on being the person who can work things out. When a tool produces plausible code in seconds, it can feel like a verdict on that skill. The feeling is a reasonable response to real change, not a flaw in you. Research on job insecurity, such as the meta-analysis by Sverke, Hellgren and Näswall (2002), has long linked the feeling of insecurity itself with poorer wellbeing, which is a good reason to deal with it deliberately instead of carrying it around.

Step one: list tasks, not "software engineer"

"Will AI replace software engineers?" is too large to answer. Two weeks of your actual work is not. Write down what you did, for example:

Mark each as production (turning a clear spec into output), judgement (deciding what to build, what could go wrong, what to leave out) or relationship (trust and communication with people). Occupational task datasets such as O*NET describe work at this level, and this is the same exercise in miniature. Tools tend to reach production work earliest. Many engineers find a large share of their value sits in the other two columns, and seeing it on paper is steadier than a general feeling.

Step two: test the tools on your own codebase

Opinions online are loud and contradictory. A small experiment is more useful. Pick two or three real tasks of different kinds, such as a bug fix, a small feature and a refactor, and try an approved tool on each. Ask three questions. Where was the output a good starting point? What did it get subtly wrong that a careful reviewer would catch? How long did checking take compared with writing it yourself?

Write the answers down and follow your employer's rules on code and data, since pasting private code into unapproved tools can be a real problem. Knowing what the tools do on your kind of work gives you something specific to say when the topic comes up in a meeting.

Step three: get good at checking

Any first draft, human or machine, needs someone who can verify it. That means knowing what correct looks like, where errors tend to hide, and how to prove a change does what it claims. Strong tests, clear review comments and a habit of reading output closely are skills that carry over directly. If your team adopts a tool, offering to write the guidelines for reviewing generated code is a practical way to shape the change instead of watching it happen.

Step four: keep a short evidence file

Once a week, write one line: what you did and what it changed. "Found that the retry logic was masking a data loss bug." "Cut the onboarding steps for new hires by rewriting the setup script." It takes two minutes. It helps at review time, when asking for a different role, and on days when the work feels interchangeable.

Step five: look sideways, calmly

You do not need to decide anything yet. It is enough to look at nearby options and talk to people in them: reliability and infrastructure, security, data work, developer tooling, or product-facing roles where you already understand the domain. Pick one, talk to two people who do it, and ask which tasks are changing and what they hire for. Career adaptability research, such as the work of Savickas and Porfeli, suggests that curiosity and concern for the future, acted on in small steps, matter more than a fixed plan.

Check your runway

Know your numbers: monthly essential costs, savings, and how long they would last if income stopped. The figure rarely matches the dread, and either way it is something you can plan around. Then protect one hour a week for learning or conversations so the plan has a place in your calendar.

Keep the worry in daylight

If thoughts about work keep you awake, write the worry down, give it a time tomorrow, and leave decisions until morning. If the worry is affecting your sleep, concentration or mood for more than a short while, talk to someone, such as a doctor, a counsellor or someone you trust. In the US, call or text 988; in the UK, Samaritans is on 116 123; elsewhere, findahelpline.com lists services by country.

A realistic next 90 days

In the first month, list your tasks, sort them, and run your three tool experiments. In the second, write down your review practices and start the weekly evidence file. In the third, speak to two people in an adjacent area and choose one small step. None of this requires you to decide whether to stay in software. It leaves you with specifics, which are easier to act on than a general dread.

Frequently asked

Will AI replace software engineers?

Nobody can say for certain. Tools are already changing parts of the work, especially production tasks such as writing code from a clear spec. Sorting your own tasks into production, judgement and relationship work shows where pressure is likely to fall in your specific job.

How can engineers show their value as coding tools improve?

Be the person who verifies and decides. Test the tools on real tasks, write review guidelines and keep a weekly evidence file of what you caught and what it changed.

What can engineers move into if they want a change?

Common adjacent areas include reliability and infrastructure, security, data work, developer tooling and domain-focused product roles. Talk to two people in one area before deciding anything.

How do I stop worrying about AI and my coding career at night?

Write the worry down, set a time in daylight to look at it, and avoid decisions in bed. If it keeps affecting your sleep or mood, speak to a health professional or a helpline.

Sources

  1. Sverke, M., Hellgren, J., & Näswall, K. (2002). No security: A meta-analysis and review of job insecurity and its consequences. Journal of Occupational Health Psychology, 7(3), 242–264.
  2. De Witte, H. (2005). Job insecurity: Review of the international literature on definitions, prevalence, antecedents and consequences. SA Journal of Industrial Psychology, 31(4), 1–6.
  3. Savickas, M. L., & Porfeli, E. J. (2012). Career Adapt-Abilities Scale: Construction, reliability, and measurement equivalence across 13 countries. Journal of Vocational Behavior, 80(3), 661–673.
  4. O*NET OnLine (U.S. Department of Labor) — occupation task data used in automation-exposure estimates; free.

The Replacement Fear

A 45-page field guide for the anxiety of being made obsolete: sleep, rumination, identity, adaptability, runway, when to get help. 150 vetted resources, four worksheets. $19 launch price, 30-day refund.

Related

Will AI take my job? Why the not-knowing is the hard partCareer adaptability is a measurable thing — and it buffers thisRunway: the one number that changes how the fear feels